Anti-Xa Activity Test Is Needed but Is Not Enough for Monitoring Fondaparinux Therapy Among Critically Ill Patients

This retrospective study included a consecutive series of patients admitted in the ICU at our hospital from February 2021 to December 2021. We collected clinical, imaging, and laboratory data in a retrospective manner. Patients were included if they received fondaparinux treatment either for thromboprophylaxis or for superficial vein thrombosis treatment. Patients were included even when they had a single anti-Xa value; they were considered to be in a stable anticoagulation state, otherwise they should have been tested for anti-Xa activity more often. In this case, the single value would be considered as the maximum value as well as the minimum value. The patients listed below were excluded: (1) patients without complete information, (2) patients without anti-Xa activity testing during fondaparinux therapy, (3) patients who already had deep venous thrombosis before fondaparinux therapy, and (4) patients who were treated with other anticoagulants during fondaparinux therapy. This study was approved by the ethics committee on biomedical research of our hospital (2021-66).

For most patients included in this study, fondaparinux was subcutaneously administered at a dose of 2.5 mg once daily, yet there were empirical dose adjustments through the whole treatment period, including but not limited to a temporary discontinuation when bleeding risk was suspected, or a decreased dose (1.5 mg) when creatinine clearance became less than 50 mL/min.

The patients in this study were divided into 3 mutually exclusive groups according to their coagulation-related outcomes.

The first group included the patients who demonstrated controlled thrombotic tendency. Thrombotic tendency was considered being under control when (1) thromboprophylaxis worked, which meant no thrombosis occurred during fondaparinux therapy (no relevant clinical symptom and no imaging evidence of thrombosis) or (2) superficial vein thrombosis (SVT) treatment worked, which meant target SVT improved or at least had no sign of therapeutic failure during fondaparinux therapy (no clinical symptom and no imaging evidence demonstrating that the size or the number of SVT increased).

The second group included the patients who demonstrated progressed thrombosis. Thrombosis was considered being progressed when (1) thromboprophylaxis failed, which meant new-onset venous thrombosis occurred during fondaparinux therapy or (2) SVT treatment failed, which meant target SVT worsened during fondaparinux therapy.

The last group included the patients who developed bleeding events. Bleeding was defined according to a consensus report from the Bleeding Academic Research Consortium (BARC).¹⁴  Once bleeding events happened to someone during fondaparinux therapy, they would be assigned to this group regardless of any other situations mentioned in the first and second groups.

Blood samples for anti-Xa activity testing were taken 3 hours after fondaparinux injection. Anti-Xa testing was performed on Sysmex CS 5100 (Sysmex Shanghai Ltd); it was calibrated against fondaparinux calibrators (TECHNOVIEW Arixtra Calibrator Set, LOT 157 2691B00, Technoclone, Vienna, Austria). Blood samples for other tests were taken according to clinical judgments but usually before breakfast. Routine coagulation testings including prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen, and antithrombin were performed on Sysmex CS 5100. Complete blood cell count was performed on Sysmex XN 9100 (Sysmex Shanghai Ltd). Thrombin-antithrombin complex (TAT) and plasmin-α2-plasmin inhibitor complex (PIC) testing were performed on Sysmex HISCL-5000 (Sysmex Shanghai Ltd). Thrombelastography (TEG) was performed on TEG 5000 (Haemonetics Corporation, Braintree, Massachusetts) using kaolin reagent. Estimated glomerular filtration rate (eGFR) testing was performed on Roche cobas c702 automatic biochemical analyzer.

Data were analyzed with Prism 9.0 software (GraphPad, San Diego, California). The Anderson-Darling test was used to assess for deviation from Gaussian distribution at the 5% significance level. Data are expressed as mean ± standard deviation for data with normal distribution, median (interquartile range) for data with nonnormal distribution, or number (percentage) for categorical data. To assess differences among groups, 1-way analysis of variance and Tukey post hoc multiple comparison tests were used for normally distributed data, a nonparametric test (Kruskal-Wallis test) was used for nonnormally distributed data, and a χ² test, Fisher exact test, or Yates continuity corrected χ² test were used for categorical data.

There were 156 anti-Xa results in total; the median value was 0.22 μg/mL (interquartile range, 0.10–0.43 μg/mL), and the minimal–maximal range was 0.01–1.41 μg/mL. Only 86 of 156 (55.1%) were within the recommended prophylactic range (0.10–0.50 μg/mL), 38 of 156 (24.4%) were less than 0.10 μg/mL, and 32 of 156 (20.5%) were greater than 0.50 μg/mL.

To identify factors that can influence anti-Xa activity of fondaparinux, we considered anti-Xa less than the first quartile as low, anti-Xa within the interquartile range as medium, and anti-Xa greater than the third quartile as high. There were 7 patients whose anti-Xa values were always low, 30 patients whose anti-Xa values were always medium, and 11 patients whose anti-Xa values were always high. Three clinical conditions and 2 laboratory indexes were different among the patients with different anti-Xa levels (Table 1). Patients who always had medium anti-Xa had the lowest rate of sepsis and vasopressor therapy. Patients who always had high anti-Xa had a highest rate of continuous renal replacement therapy. Patients who always had low anti-Xa had the lowest antithrombin level and a medium eGFR. Patients who always had medium anti-Xa had a medium antithrombin level but the highest eGFR. Patients who always had high anti-Xa had the highest antithrombin level, but the lowest eGFR. The remaining 22 patients could not be classified into above-mentioned categories. For example, 1 patient had 2 anti-Xa values: 1 was low, the other one was high. Therefore their data were not analyzed in this part.

There were 70 patients included in total. Thrombotic tendency was controlled in 32 of 70 patients (45.7%), thrombosis progressed in 22 of 70 patients (31.4%), and bleeding events occurred in 16 of 70 patients (22.9%). Patients with bleeding events were found to have more vasopressor therapy and a worse prognosis (more death and more treatment withdrawal) (Table 2). The patients with progressed thrombosis had 54 anti-Xa values; 17 of 54 (31.5%) were less than 0.10 μg/mL, even though this proportion was twice that of patients with controlled thrombotic tendency (11 of 72, 15.3%), it was similar to that of patients with bleeding events (10 of 30, 33.3%). The proportions of patients with anti-Xa greater than 0.43 μg/mL, and maximum and minimum anti-Xa values, were comparable among the 3 groups (Table 2). Neither maximum anti-Xa level nor minimum anti-Xa levels during fondaparinux therapy were different among the 3 groups (Table 2).

PT, APTT, fibrinogen, and platelet counts were not different among patients with different coagulation-related outcomes. PIC and TEG kinetics time values, angle α value, and maximum amplitude values were not different either.

TAT was higher in patients with progressed thrombosis compared to patients with controlled thrombotic tendency (Figure 1, A and B; Table 3). There were 30 patients who had TAT continuously tested, including 9 patients with controlled thrombotic tendency, 15 patients with progressed thrombosis, and 7 patients with bleeding events. Among patients with controlled thrombotic tendency, there was a gradual decline in TAT value (Figure 1, C and D), and they all had a TAT value less than 8 ng/mL (twice the upper reference value) after fondaparinux treatment for 1 week (Figure 1, C and D). Among patients with progressed thrombosis, TAT decreased at the beginning of fondaparinux treatment, but would rebound around 5 to 7 days later (Figure 1, C and E). No particular pattern was observed in patients with bleeding events (Figure 1, C and F).

The TEG reaction time (R) value was significantly higher in patients with bleeding events compared to patients with progressed thrombosis, as well as compared to patients with controlled thrombotic tendency, but did not reach statistical significance (Figure 2, A; Table 3). There were 28 patients who had TEG continuously monitored, including 9 patients with controlled thrombotic tendency, 13 patients with progressed thrombosis, and 6 patients with bleeding events. Among patients with bleeding who had TEG continuously monitored, 66.7% (4 of 6) had at least 1 R value above the upper reference value; this proportion was more than 3 times that of patients without bleeding (18.2%, 4 of 22) (P = .01) (Figure 2, B though D).

As expected, antithrombin level and creatinine clearance were the main factors that influenced anti-Xa activity of fondaparinux. Antithrombin deficiency resulted in low anti-Xa levels,^15–17  while impaired creatinine clearance, which indicated renal insufficiency, led to high anti-Xa levels. Unfortunately, a critically ill patient can experience antithrombin deficiency and renal insufficiency at the same time, which is one of the reasons that fondaparinux is unpredictable among the critically ill.

In addition, this study found that vasopressor therapy was related to the uncertainty of fondaparinux response in the ICU population. The proportion of patients who received vasopressor therapy was lower among the patients with medium anti-Xa, whether compared to the patients with low anti-Xa or compared to the patients with high anti-Xa. This relationship between vasopressor therapy and the uncertainty of fondaparinux response can be explained by the patients’ conditions and by the vasopressors’ pharmacologic actions. Firstly, patients who need vasopressor therapy usually have critical conditions such as multiple organ failure that can cause unpredictable pharmacokinetic disturbances. Secondly, vasopressors themselves can disturb pharmacokinetic processes of fondaparinux. On one hand, vasopressors can cause adrenergic vasoconstriction of peripheral blood vessels, leading to impaired perfusion of subcutaneous tissue and therefore impaired bioavailability of subcutaneously administered fondaparinux.¹⁰  On the other hand, vasopressors might induce renal hypoperfusion, leading to impaired excretion of fondaparinux. There was 1 study in the literature also demonstrating a trend toward higher levels of anti-Xa of patients on vasopressor therapy when compared to hemodynamically stable patients.³  Another factor related to the uncertainty of fondaparinux response in the ICU population was the concurrence of sepsis. The proportion of patients who suffered sepsis was smaller among the patients with medium anti-Xa, whether compared to the patients with low anti-Xa, or compared to the patients with high anti-Xa. This is probably because sepsis patients also have critical conditions that might disturb the pharmacokinetic processes of fondaparinux; besides, in this study, 38 of 43 (88.4%) sepsis patients were on vasopressor therapy.

The above-mentioned results suggested that, in the ICU population, fondaparinux dose-response effect was unpredictable due to complicated medical conditions. This can probably explain why only half of the anti-Xa values were within the recommended prophylactic range in this study. In consideration of this unpredictable dose-response effect and the fragile balance of thrombotic to bleeding risk in the ICU population, we suggest routinely testing anti-Xa activity to monitor the fondaparinux drug-response effect in the ICU population.

However, although anti-Xa activity testing might be enough to help guide anticoagulation therapy among patients with simple medical conditions, such as VTE after planned surgery, it did not appear to translate to the risk of thrombosis or bleeding very well during fondaparinux therapy among the critically ill in the present study. Usually, a low anti-Xa value would indicate the risk of thrombosis, and a high anti-Xa value would indicate the risk of bleeding. However, in this study, compared to patients with controlled thrombotic tendency, both patients with progressed thrombosis and patients with bleeding were twice as likely to demonstrate a low anti-Xa value (Table 2). In other words, a low anti-Xa value could be translated to the risk of either thrombosis or bleeding, which could confuse clinicians on decision making, because the treatments of thrombosis and bleeding can be totally different. Nevertheless, this paradox can be explained. Whether there is thrombin generation is based on the balance between anticoagulation and coagulation. However, the coagulation status of critically ill patients is extremely variable because of their complicated medical conditions. For example, they might have lower coagulation factors due to liver dysfunction^13,18 ; they might also have higher factor VIII due to an acute response phase¹⁹ ; they might be treated with fresh frozen plasma or human prothrombin complex concentrate frequently; and they might be treated with an artificial liver support system or continuous renal replacement therapy. Under such circumstances, anti-Xa itself, which only represents anticoagulation, is no longer enough to be related to the risk of thrombosis or bleeding. Therefore, anti-Xa activity testing is less helpful in guiding anticoagulation therapy among the critically ill.

Other than the balance between anticoagulation and coagulation, the balance between thrombin generation and plasmin generation is also of great importance for thrombogenesis. We therefore analyzed TAT and PIC values, which can represent thrombin generation and plasmin generation, respectively.²⁰  In the present study, PIC values were slightly higher than the upper reference value in all groups, and they were comparable to each other. In contrast, TAT values representing different degrees of thrombin generation showed a promising predictive value for clinical outcomes. When compared to patients with controlled thrombotic tendency, patients with progressed thrombosis had statistically significant higher TAT levels. Patients with bleeding events also had higher TAT levels, but these were not statistically significant. Moreover, in patients with controlled thrombotic tendency, TAT gradually decreased to a level less than 8 ng/mL (twice upper reference value) after 1 week of fondaparinux treatment, but in patients with progressed thrombosis, TAT rebounded in around 5 to 7 days after fondaparinux treatment even though it decreased at the very beginning. These results indicated that a low TAT value could reflect a low risk of VTE, and continuous TAT monitoring might help to assess whether anticoagulation treatment is effective or not.

In the present study, major bleeding (BARC level 5) occurred in 1 patient, and moderate bleeding (BARC level 3) occurred in 10 patients (total incidence, 15.7% [11 of 70]); this was higher than prophylactic low-molecular-weight heparin or unfractionated heparin in the ICU population (5.5–5.6%).²¹  Neither routine coagulation testing including PT, APTT, and fibrinogen nor advanced hemostatic testing including TAT and PIC were helpful to assess the risk of bleeding during fondaparinux therapy in this study. Platelet count was not statistically different between patients with or without bleeding, either. We found that the TEG value was different between patients with or without bleeding. The TEG R value was higher in patients with bleeding events compared to patients with progressed thrombosis, or compared to patients with controlled thrombotic tendency. Besides, a TEG R value above the upper reference value occurred more frequently among patients with bleeding compared to patients without bleeding, indicating TEG could be helpful to identify bleeding risk during fondaparinux therapy.

This study has a few limitations. Firstly, this is a retrospective study, and the patients were from a single center. Secondly, there were only 70 patients with 156 anti-Xa values being included in this study. The sample size was small because clinicians do not usually have their fondaparinux-treated patients tested for anti-Xa activity, as they are familiar with the recommendation provided by the manufacturer, which states that fondaparinux does not need to be routinely monitored, even though this recommendation is not based on data generated from critically ill patients. We hope our study will draw attention to fondaparinux monitoring among critically ill patients, so that well-designed clinical studies can be carried out to gain more information about fondaparinux therapy in the future.

In conclusion, the fondaparinux dose-response effect among critically ill patients was unpredictable; it is very important to monitor fondaparinux therapy among them. In our study, we demonstrated that anti-Xa activity, TAT, and TEG testing were most helpful to guide precise fondaparinux therapy in the ICU.